Associative recognition in Alzheimer's disease: evidence for impaired recall-to-reject.

Patients with mild Alzheimer's disease (AD) were compared with age-matched control subjects on an associative recognition task. Subjects studied pairs of unrelated words and were later asked to distinguish between these same studied pairs (intact) and new pairs that contained either rearranged studied words (rearranged) or non-studied words (non-studied). Studied pairs were presented either once or 3 times. Repetition increased hits to intact pairs in both groups, but repetition increased false alarms to rearranged pairs only in patients. This latter pattern indicates that repetition increased familiarity of the rearranged pairs, but only the control subjects were able to counter this familiarity by recalling the originally studied pairs (a recall-to-reject process). AD impaired this recall-to-reject process, leading to more familiarity based false alarms. These data support the idea that recollection-based monitoring processes are impaired in mild AD.